vim-patch:8.1.1122: char2nr() does not handle composing characters

Problem:    char2nr() does not handle composing characters.
Solution:   Add str2list() and list2str(). (Ozaki Kiichi, closes vim/vim#4190)
9d40128afd

'utf8' optional param is noop unlike Vim.

src/nvim/eval.lua

@@ -214,6 +214,7 @@ return {
     line={args=1},
     line2byte={args=1},
     lispindent={args=1},
+    list2str={args={1, 2}},
     localtime={},
     log={args=1, func="float_op_wrapper", data="&log"},
     log10={args=1, func="float_op_wrapper", data="&log10"},
@@ -321,6 +322,7 @@ return {
     sqrt={args=1, func="float_op_wrapper", data="&sqrt"},
     stdpath={args=1},
     str2float={args=1},
+    str2list={args={1, 2}},
     str2nr={args={1, 2}},
     strcharpart={args={2, 3}},
     strchars={args={1,2}},

src/nvim/eval/funcs.c

@@ -5246,6 +5246,36 @@ static void f_lispindent(typval_T *argvars, typval_T *rettv, FunPtr fptr)
   }
 }
 
+// "list2str()" function
+static void f_list2str(typval_T *argvars, typval_T *rettv, FunPtr fptr)
+{
+  garray_T ga;
+
+  rettv->v_type = VAR_STRING;
+  rettv->vval.v_string = NULL;
+  if (argvars[0].v_type != VAR_LIST) {
+    EMSG(_(e_invarg));
+    return;
+  }
+
+  list_T *const l = argvars[0].vval.v_list;
+  if (l == NULL) {
+    return;  // empty list results in empty string
+  }
+
+  ga_init(&ga, 1, 80);
+  char_u buf[MB_MAXBYTES + 1];
+
+  TV_LIST_ITER_CONST(l, li, {
+    buf[utf_char2bytes(tv_get_number(TV_LIST_ITEM_TV(li)), buf)] = NUL;
+    ga_concat(&ga, buf);
+  });
+  ga_append(&ga, NUL);
+
+  rettv->v_type = VAR_STRING;
+  rettv->vval.v_string = ga.ga_data;
+}
+
 /*
  * "localtime()" function
  */
@@ -9353,6 +9383,17 @@ static void f_str2float(typval_T *argvars, typval_T *rettv, FunPtr fptr)
   rettv->v_type = VAR_FLOAT;
 }
 
+// "str2list()" function
+static void f_str2list(typval_T *argvars, typval_T *rettv, FunPtr fptr)
+{
+  tv_list_alloc_ret(rettv, kListLenUnknown);
+  const char_u *p = (const char_u *)tv_get_string(&argvars[0]);
+
+  for (; *p != NUL; p += utf_ptr2len(p)) {
+    tv_list_append_number(rettv->vval.v_list, utf_ptr2char(p));
+  }
+}
+
 // "str2nr()" function
 static void f_str2nr(typval_T *argvars, typval_T *rettv, FunPtr fptr)
 {

src/nvim/testdir/test_utf8.vim

@@ -60,3 +60,46 @@ func Test_getvcol()
   call assert_equal(2, virtcol("'["))
   call assert_equal(2, virtcol("']"))
 endfunc
+
+func Test_list2str_str2list_utf8()
+  " One Unicode codepoint
+  let s = "\u3042\u3044"
+  let l = [0x3042, 0x3044]
+  call assert_equal(l, str2list(s, 1))
+  call assert_equal(s, list2str(l, 1))
+  if &enc ==# 'utf-8'
+    call assert_equal(str2list(s), str2list(s, 1))
+    call assert_equal(list2str(l), list2str(l, 1))
+  endif
+
+  " With composing characters
+  let s = "\u304b\u3099\u3044"
+  let l = [0x304b, 0x3099, 0x3044]
+  call assert_equal(l, str2list(s, 1))
+  call assert_equal(s, list2str(l, 1))
+  if &enc ==# 'utf-8'
+    call assert_equal(str2list(s), str2list(s, 1))
+    call assert_equal(list2str(l), list2str(l, 1))
+  endif
+
+  " Null list is the same as an empty list
+  call assert_equal('', list2str([]))
+  " call assert_equal('', list2str(test_null_list()))
+endfunc
+
+func Test_list2str_str2list_latin1()
+  " When 'encoding' is not multi-byte can still get utf-8 string.
+  " But we need to create the utf-8 string while 'encoding' is utf-8.
+  let s = "\u3042\u3044"
+  let l = [0x3042, 0x3044]
+
+  let save_encoding = &encoding
+  " set encoding=latin1
+
+  let lres = str2list(s, 1)
+  let sres = list2str(l, 1)
+
+  let &encoding = save_encoding
+  call assert_equal(l, lres)
+  call assert_equal(s, sres)
+endfunc